RESEARCH / Focused target
Connected graphs with ordinary domination root minus four
RELEASED FOUNDATIONS / PROPOSED CONNECTION
Research connections
All source anchors (1)
- D5/S3/Combinatorics/Graph/Domination/MinusFourReassembly.resultAbsent from this graph
Problem
The question
Saeid Alikhani and Max Griswold, On the Integer Domination Root Conjecture, arXiv:2608.00109v1, Section 4 item 3 asks:
Does there exist an infinite family of graphs G satisfying D(G, -4) = 0?
The source uses finite simple undirected connected graphs. For ordinary domination, a subset S dominates when each vertex belongs to S or has a neighbor in S. The polynomial counts each such subset exactly once:
$$D(G,x)=\sum_{S\text{ dominates }G}x^{|S|}.$$
The exact result is: for every natural N there exist n>N and a simple graph G on Fin n such that G is connected and D(G,-4)=0. Unbounded orders ensure infinitely many isomorphism classes. Connectedness applies to G; the dominating subsets need not themselves be connected. There is no additional degree, planarity, cyclomatic-number, or root-multiplicity hypothesis.
The declaration is
D5/S3/Combinatorics/Graph/Domination/MinusFourReassembly.result.
Its ordinary dominating subsets are exactly DominatingSetAverage.domSets;
the summands are integer powers of the integer minus four. Both existential
witnesses follow the universal N, and the inequality is strict.
Motivation
Our foothold
The source's Theorem 2.1 already supplies a graph G33 with 33 vertices and 36 edges whose domination polynomial has a factor x+4. That finite example is prior art. The question requires a construction of unbounded order. The existing decreasing-predecessor connectedness theorem supplies paths in the base graph without a second connectivity proof or a renamed supplier.
Gap
Missing bridges
The infinite-family conclusion follows from a root-preserving support replication operation. The result does not assert that the chosen base is smallest, classify all connected graphs with this root, establish any other integer root, or constrain the family's cycle rank or maximum degree. These are independent questions outside the delivered statement.
Route
Proposed approach
The general Decomposition interface permits arbitrary dimensions n,k,m,
dependent branch sizes, a graph, vertex labels, roots, incidence data and an
integer weight x. It imposes the exact local neighbor and leaf conditions.
Restriction to the core and branch subsets is inverse to disjoint assembly.
Cardinality adds, ordinary domination becomes internal branch domination
plus coverage of the core, and the weighted sum factors into exact
root-conditioned branch messages. This proves reassembly; substituting a
table whose entries are those same messages gives zero_of_messages.
The actual application uses G33, the four-cycle core 0,1,2,3, and eight branches of sizes 2,2,2,2,6,3,3,9. At x=-4, the selected-root messages are (12,12,12,12,972,-36,-36,-20736); the omitted-root messages are (-4,-4,-4,-4,-1024,16,16,5888). The exact same decomposition supplies the graph laws, message equalities and covered-core cancellation. None of these numeric premises is left assumed. The resulting evaluation of G33 is zero.
Vertex 5 has exactly vertex 4 as its neighbor. For r copies of G33, join each copied support 4 to a new hub and attach one pendant to the hub. The weighted evaluation of the resulting graph is
$$x(x+2)D(G33,x)^r.$$
At x=-4 this vanishes for every r>=1. Paths through each support and the hub prove connectedness. The graph has 33r+2 vertices, so r=N+1 gives 33(N+1)+2>N. Transport by a finite equivalence gives the required Fin n vertex type. The case r=0 is K2 and its value at -4 is 8; it is not a zero-valued member of this family.
Falsifier
What would falsify this route
A mismatch between the ordinary dominating-subset sum and a connected-subset variant, an unproved table entry, a failed graph law, a disconnected family member, or a bounded-order construction would fail this exact statement. The finite base alone would not settle it. The general checker requires its message table and cancellation to refer to the same graph and weight.
Evidence
Evidence to collect
The exact universal theorem, its general reassembly checker and its actual application have current scoped Lean kernel evidence with only Classical.choice, Quot.sound and propext, and no sorryAx. Both retained Reg sources compile; compiled family evidence is distinct from completed four-slot registration. The root and all dimensions in the public formulas match the typed Lean statements.
The finite G33 and its reported root are attributed to Theorem 2.1 of the source. The restriction/reassembly argument and support-family application are repository synthesis. The note cites v1 and retains citation-only use of the arXiv source; no author code or substantial source prose is reproduced. The licensed source material is not assigned an invented Creative Commons or publisher license.
Triage
Scope assessment
This is the complete unbounded connected-family result, not a finite-order
test or a result about connected dominating subsets. Its external
open-problem-resolution basis is issue
#25087.
The result's exact statement identifier is
sha256:faa4a52a3fd8fb6f55ca09475a596c68a98037e568e5045f172a5e46bb6a4eb1.
Every auxiliary theorem in the new host is used on the derivation path;
the family result supplies the external settlement basis without claiming
an independent escape witness for those organizational helpers.
ASSUMED-UNVERIFIED
Unverified assumptions
The bounded prior source screen found no exact later resolution on its checked surfaces. Literature completeness and worldwide priority remain ASSUMED-UNVERIFIED; absence in that screen does not establish global nonresolution or originality.
Four public escape audits remain unfinished and disclosed at
#25087, comment 6553848217:
MinusFourReassembly.reassembly, MinusFourReassembly.zero_of_messages,
MinusFourReassembly.result, and the original
ConsecutiveCycleSumsGraph.connected_of_predecessor. The first two have
compiled typed family evidence; the prior result-host Contract discovery
failed with contract.discovery:rigid_universes at registration_1, and
the second binding was not reached. The result retains its unfinished
exact bridge/family/Contract boundary. The supplier has compiled full
telescope family evidence but no Contract wrapper; no rigid-universes
failure is attributed to it. Lawful enrollment, enrolled readout,
sourceSelection reconstruction and current binding evidence are missing;
none is declared_validated. The existing linked-issue exception permits
delivery while keeping these obligations explicit.